Open Access
Add to BibSonomy
Abstract
We propose an optical line terminal (OLT)-centralized sampling frequency offset (SFO) compensation scheme for adaptively-modulated OFDM-PON systems. By using the proposed SFO scheme, the phase rotation and inter-symbol interference (ISI) caused by SFOs between OLT and multiple optical network units (ONUs) can be centrally compensated in the OLT, which reduces the complexity of ONUs. Firstly, the optimal fast Fourier transform (FFT) size is identified in the intensity-modulated and direct-detection (IMDD) OFDM system in the presence of SFO. Then, the proposed SFO compensation scheme including phase rotation modulation (PRM) and length-adaptive OFDM frame has been experimentally demonstrated in the downlink transmission of an adaptively modulated optical OFDM with the optimal FFT size. The experimental results show that up to ± 300 ppm SFO can be successfully compensated without introducing any receiver performance penalties.
© 2017 Optical Society of America
1. Introduction
Optical orthogonal frequency-division multiplexing (OOFDM) is a promising modulation format for the next-generation passive optical network (PON) because of its high spectral efficiency (SE) and the resistance to optical fiber dispersions such as chromatic dispersion and polarization mode dispersion [1,2]. It has been extensively studied by both off-line experiments and real-time digital signal processing (DSP) approaches [1–10]. However, OOFDM systems are very sensitive to frequency offsets, i.e., carrier frequency offset (CFO) and sampling frequency offset (SFO). Fortunately, there aren’t CFOs existing in the Hermitian symmetry (HS)-based intensity-modulated and direct-detection OFDM (IMDD-OFDM) systems which have been considered as a preferred candidate for future cost-sensitive PON due to its simple structure and low-cost compared to coherent optical OFDM [4].
An SFO between the digital-to-analog converter (DAC) in the transmitter and the analog-to-digital converter (ADC) in the receiver would severely degrade the transmission performance. The effects of SFO on the OFDM signals can be characterized by phase rotation and inter-channel interference (ICI) in the frequency domain [10], and timing offset of the fast Fourier transform (FFT) window in the time domain [11]. On the one hand, the ICI is small and can be regarded as additional noise when the SFO is not so large. On the other hand, it may introduce serious inter-symbol interference (ISI), when the amount of timing offset is larger than the applied cyclic prefix (CP) or cyclic suffix (CS) length [12]. Therefore, the SFO compensation is mainly to correct phase rotations and avoid SFO-induced ISI.
In the literature, there are several methods have been proposed and demonstrated in IMDD-OFDM systems. The first one is the CP-based SFO compensation [13], where the estimated SFO is used to control the ADC clock frequency via an external voltage controlled oscillator (VCO). It needs high-precision and stable VCO and additional circuits; the second one is to transmit a dedicated clock at the transmitter side [14], which requires more complex circuits; the third one is to estimate the SFO-induced phase rotations using known pilot symbols in each data-carrying OFDM symbol and then compensate SFO [15–17], which results in reduced SE. The last one is to use the training symbols (TSs) at the beginning of every OFDM frame to estimate and compensate the SFO [18]. It improves SE and estimation accuracy in the case of a small range of SFO ( ± 60 ppm). However, a standard SFO of up to 200 ppm should be compensated in an optical OFDM system [12, 19].
This paper is an extension of our early work presented at ACP 2016 [20], and only phase rotation modulation (PRM) was used to correct the SFO-induced phase rotations in an end-to-end IMDD-OFDM transmission system in [20]. In this work, an efficient optical line terminal (OLT) centralized SFO compensation scheme is proposed for OFDM-PON systems. On the one hand, PRM and phase compensation are applied to correct SFO-induced phase rotations existing in downstream transmitter and upstream receiver, respectively. On the other hand, length-adaptive OFDM frame is used to avoid SFO-induced ISI. Compared to the traditional pilot-aided SFO compensation scheme which is implemented in ONUs, such an SFO compensation scheme can reduce the complexity of optical network units (ONUs) as well as maintain a high SE. The optimal fast Fourier transform (FFT) size is first identified in the IMDD-OFDM system in the presence of an SFO. The proposed SFO compensation scheme including PRM and length-adaptive OFDM frame has been experimentally demonstrated in the downlink transmission of an adaptively modulated optical OFDM with the optimal FFT size.
2. Operation principles of the proposed OLT-centralized SFO compensation scheme
The OFDM-PON with the proposed OLT-Centralized SFO compensation scheme is schematically depicted in Fig. 1. In the initial stage of establishing connections between OLT and ONUs, a probe OFDM signal for SFO estimation generated by OLT is first sent to all ONUs, and then the SFOs between OLT and each ONU are estimated, and then sent back to OLT via the uplink for the correction of SFO-induced phase rotations (i.e., PRM) for downstream transmitter and phase compensation for upstream receiver). Here, SFO can be regarded as one type of channel status information (CSI). Thus, the proposed SFO compensation scheme is compatible with adaptive modulation technique. Additionally, the lengths of cyclic prefix/cyclic suffix (CP/CS) and/or the number of OFDM symbols in every OFDM frame also can be adaptively adjusted according to the estimated SFOs. In this way, the SFO-induced ISI can be effectively avoided.
2.1 SFO estimation, PRM and phase compensation
Generally, a data-carrying OFDM frame consists of several training symbols and followed by data-carrying OFDM symbols. The timing synchronization and channel equalization are usually performed with the help of these training symbols. Here, a probe OFDM signal has a similar structure with the data-carrying OFDM frame, which consists of one training symbol and NT QPSK-OFDM symbols, is applied to estimate SFO. The theoretical SFO between OLT and the n-th ONU is defined as, where (nominal frequency) and are the clock frequencies for OLT and the n-th ONU, respectively. Once timing synchronization and channel equalization are done, the estimated SFO, , can be estimated by using a similar method as described in [11, 15] and expressed as
where where NF is FFT size, NS = NF + NCP + NCS is the length of an OFDM symbol with CP and CS, and ND denotes the number of the data subcarriers (SCs) in the positive frequency bins. represents the SFO-induced phase rotation on the k-th SC of the m-th OFDM symbol.Only the parameter Sn is sent back to OLT for PRM and phase compensation, i.e., the mapped data or equalized data on the k-th SC of the m-th OFDM symbol will be multiplied by a factor assuming that the k-th SC has been assigned to the n-th ONU.
2.2 Length-adaptive CP/CS and OFDM frame
To avoid the SFO-induced ISI, the FFT windows of the received OFDM samples are dynamically adjusted by monitoring SFO-induced phase rotation in the frequency domain [12]. Alternately, length-adaptive CS/SP and OFDM frame according to the estimated SFO can also be employed to avoid the SFO-induced ISI. The SFO-induced timing offset of the FFT windows for the last data-carrying OFDM symbol in every OFDM frame can be defined as. If the NCP, NCS and NT of each OFDM frame have the following relationship
then SFO will not introduce ISI. However, increasing CP/CS length will result in low SE. Only length-adaptive OFDM frame (NT) is experimentally verified in this work.3. Experimental setup
The experimental setup of adaptively modulated IMDD-OFDM downstream transmission with the proposed SFO compensation scheme is illustrated in Fig. 1. At the transmitter side, the digital OFDM signal is generated offline with the transmitter DSP in Matlab, where the IFFT size (NF) is 32/64/128/256/512/1024/2048, and the corresponding CP and CS length (NCP/NCS) are 1/32 IFFT size. The number of positive-frequency SCs (ND) which assigned to carry data is 13/26/52/104/208/416/832. The input vector to the IFFT is constrained to have Hermitian symmetry (HS) for the real-valued OFDM signal generation. Only one TS is utilized to realize both timing synchronization and zero-forcing channel estimation. The offline generated signal is digitally clipped at a clipping ratio of 12.5 dB before being sent to an arbitrary waveform generator (AWG, Tektronix AWG7122C) with DAC working at 10-bit and 20 GS/s sample rate (RS). The OFDM frame consists of 1 TS and followed by 100 data-carrying OFDM symbols (NT = 100). So the bandwidth of the OFDM signal is ND/NF*RS = 8.125 GHz for different IFFT size cases. The converted signal is first amplified by a 14 GHz electrical amplifier (EA) and then drives a 10 Gb/s distribute feedback (DFB)-based directly-modulated laser (DML). The intensity-modulated optical double-sideband (ODSB) signal with a launch power of 3 dBm is transmitted over 10 km SMF-28 and then directly detected by 10 GHz AC-coupled photo-detector (PD) at the receiver. Then, the received baseband signal is sampled by a digital storage oscilloscope (DSO, Teledyne Lecroy Wavemaster 820Zi-A) with ADC operating at 8-bit and 40 GS/s. Subsequently, the DSO captured samples are further processed offline with the receiver DSP, where the intra-symbol frequency averaging (ISFA) technique [21] is applied to obtain high-accuracy channel estimation, and the optimal ISFA length is used in our experiments. To emulate the desired SFO between DAC and ADC, the sampling rate of the AWG is appropriately changed. It should be noted that the separate reference clocks are used for the AWG and DSO. The electrical spectrum of the received OFDM signal with 256-point IFFT and optical spectra of the received optical carrier and ODSB-OFDM signal are inserted in Figs. 2(a) and 2(b), respectively. Here, a peak at ~15 GHz in the electrical spectrum should be the ADC sampling clock noise [22].
Fig. 2 Experimental setup of adaptively modulated IMDD-OFDM downstream transmission with the proposed SFO compensation scheme.
It is important to identify the optimal IFFT/FFT size when SFO exists before we experimentally investigated the performance of the proposed SFO compensation method. It also should be pointed that two probe QPSK-OFDM signals without and with PRM are sent by the transmitter to estimate SFO for PRM and signal-to-noise ratio (SNR) for adaptive modulation, respectively, in the initial stage of establishing a connection over the SMF-28 link. For simplicity’s sake, only bit loading is performed with look-up table (LUT) method according to the estimated SNRs.
4. Results and discussions
4.1 FFT size optimization
To identify the optimal FFT size, two QPSK-OFDM frames without and with PRM under SFO free and 200 ppm SFO, respectively, are sent by the transmitter. After 10 km SMF-28 transmission, the error vector magnitudes (EVMs) in dB as a function of FFT size are shown in Fig. 3. When SFO is free, the EVM performance is improved with the increase of FFT size. It shows that there is about 4 dB EVM improvement when the FFT size increases from 32 to 2048. This fact is due to that the OFDM signal with large FFT size has less side-lobe power leakage and thus is less sensitive to narrow filtering effects [23-24]. For the SFO of 200 ppm and PRM enabled case, the EVM performance of the OFDM signal with large FFT size is degraded, which is mainly attributed to the SFO-induced severe ICI. Therefore, the optimal FFT size for our experimental transmission system is 256. It should be pointed that there isn’t SFO-induced ISI for each FFT size according to the expression (4).
4.2 SFO estimation and PRM performance
The SFO estimation performance after 10 km SMF-28 transmission is shown in Fig. 4(a), where the estimated SFO is a function of initial SFO. It shows that a wide range of SFO from −1000 ppm to 1000 ppm can be accurately estimated by using the proposed SFO estimation method, and the estimation accuracy depends on the number of OFDM symbols (i.e., NT). When NT is 40, the best estimation performance with an accuracy of ± 1 ppm for SFO from −1000 ppm to 1000 ppm can be observed from Fig. 4(b). In the large SFO cases, the SFO-induced ICI and ISI may reduce the accuracy of SFO estimation when NT is large enough, while a variety of noises may be the reason of the reduced estimation accuracy for the small NT cases such as NT = 20.
The measured EVM performance as a function of SFO is plotted in Fig. 5. It shows clearly that the EVM penalties are negligible when SFO in the range of −300 ppm to 300 ppm by only using the proposed PRM method. Two QPSK constellations under SFO-free and 300 ppm are given in Figs. 5(c) and 5(d), respectively. In the case of without PRM, the EVM performance with the increase of SFO is dramatically degraded, which is mainly due to the increasing phase rotation, ISI and ICI caused by SFO. The corresponding QPSK constellations under −100 and −1000 ppm without PRM are inserted in Figs. 5(a) and 5(b), respectively. For the PRM-enabled cases, the SFO-induced ISI and ICI are the main reasons of the degraded EVM performance. The corresponding constellation is inserted in Fig. 5(e).
Fig. 5 Measured EVM performance versus SFO and QPSK constellations: (a) w/o PRM, −1000 ppm; (b) w/o PRM, −100 ppm; (c) w/o PRM, w/o SFO; (d) w/ PRM, 300 ppm and (e) w/ PRM, 1000 ppm (NT = 100, NF = 256, NCP/NCS = 8).
4.3 Performance improvements with length-adaptive OFDM frame
In addition to the PRM, the length-adaptive OFDM frame can be used to avoid the SFO-induced ISI and then further improve the transmission performance. The EVM performance is dramatically deteriorated as the increase of SFO and NT in the case of without PRM as shown in Fig. 6(a). For the PRM-enabled cases, the EVM performance can be optimized for the system with different SFOs by adaptively adjusting the length of OFDM frame. As shown in Fig. 6(b), there are about 0.5, 3 and 4 dB EVM penalties when NT is 100 in the presence of 400, 800 and 1000 ppm of SFOs, respectively. According to the expression (4) mentioned above, the maximize NT should be about 72, 35 and 28. By doing so, the SFO-induced ISI will be avoided. As a result, the EVM penalties are only 0.5 and 1 dB under 800 and 1000 ppm of SFOs, respectively, while the EVM penalty is ignored at 400 ppm SFO.
The SE reduction as a function of the amount of SFO due to the use of the proposed length-adaptive OFDM frame is shown in Fig. 6(c). It exhibits that the SE reduction is less than 3.5% and can be neglected when the SFO in a large range from −1000 to 1000 ppm.
4.4 Transmission performance of adaptively modulated IMDD-OFDM with PRM
For simplicity’s sake, bit loading (BL) is only performed in this work, which is based on the relationship between required SNR and desired BER at 1 × 10−3 in different QAM modulations [25]. The channel SNR can be calculated from EVMs of individual SCs of the probe OFDM signal [26] and the adaptively loaded number of bits according to the channel SNR, are shown in Fig. 7(a) and 7(b), respectively.
The OFDM frame consists of 1 TS and followed by 100 adaptively modulated OFDM symbols. Thus, the net bit rate RN can be calculated by
where Nb represents the number of information-bearing bits per OFDM symbol, which is 382 in our experiment, and OHFEC donates the FEC overhead. Assume that 7% the hard-decision FEC (HD-FEC) is used to achieve error-free transmission for the pre-BER of 1 × 10−3, and then RN = (382 × 100 × 20e9)/((100 + 1) × (256 + 8 + 8))/(1 + 0.07) b/s = 26 Gb/s excluding the overheads of TS, CP/CS and 7% FEC code.After 10 km SMF-28 transmission, the received optical power (ROP) is fixed at 0 dBm and the SFOs are initialized to −200, 0, 200 ppm, respectively, and the EVM and BER performances of the 26 Gbit/s adaptively modulated and PRM-enabled OFDM signals over 45 minutes are measured and shown in Figs. 8(a) and 8(b). It shows that the measured BERs are below the desired BER of 1 × 10−3 and EVMs of less than −14 dB are stable over the measurement period. The corresponding constellations are also shown in Fig. 9. It indicates that the SFO-induced phase rotation can be successfully corrected by using the proposed PRM method. Note that the NT is set to 100 which is less than the maximum allowable value of NT, 146, for ± 200 ppm SFOs. Therefore the ISI induced by SFO can be avoided in our experiments.
Fig. 9 128/64/32/16/8/4/2-QAM constellations of the adaptively-modulated OFDM signal with PRM (SFO = 200 ppm).
5. Conclusions
We have proposed a novel OLT-centralized SFO compensation scheme for OFDM-PON. On the one hand, the SFO-induced phase rotation can be effectively corrected by using PRM and phase compensation for the downstream transmitter and upstream receiver, respectively; on the other hand, the length-adaptive OFDM frame can be used to avoid the SFO-induced ISI. The PRM and length-adaptive OFDM have been experimentally verified in an adaptively-modulated end-to-end IMDD-OFDM transmission system. The results indicated that the phase rotations caused by up to ± 300 ppm SFO could be successfully corrected with negligible EVM penalties. Moreover, large SFO-induced heavy ISI can be avoided by using the length-adaptive OFDM frame. Also, the good EVM and BER stabilities have also been observed over the measurement period. We believe that the proposed SFO compensation scheme has great application penitential in adaptively modulated optical OFDM systems.
Funding
Hunan Provincial Natural Science Foundation of China (2017JJ3212, 2016JJ6097, 14JJ6007); Scientific Research Fund of Hunan Provincial Education Department (17C0957, 14B119); Project Supported for excellent talents in Hunan Normal University (ET1502).
References and links
1. A. J. Lowery, D. Liang, and J. Armstrong, “Orthogonal frequency division multiplexing for adaptive dispersion compensation in long haul WDM systems,” in OFC (OSA, 2006), paper PDP39.
2. I. B. Djordjevic, “PMD compensation in fiber-optic communication systems with direct detection using LDPC-coded OFDM,” Opt. Express 15(7), 3692–3701 (2007). [CrossRef] [PubMed]
3. J. Xiao, Z. Zhang, X. Li, Y. Xu, L. Chen, and J. Yu, “OFDM vector signal generation based on optical carrier suppression,” IEEE Photonics Technol. Lett. 27(23), 2449–2452 (2015). [CrossRef]
4. Y. Gao, J. Yu, J. Xiao, Z. Cao, F. Li, and L. Chen, “Direct-detection optical OFDM transmission system with pre-emphasis technique,” J. Lightwave Technol. 29(14), 2138–2145 (2011). [CrossRef]
5. J. Xiao, X. Li, Y. Xu, Z. Zhang, L. Chen, and J. Yu, “W-band OFDM photonic vector signal generation employing a single Mach-Zehnder modulator and precoding,” Opt. Express 23(18), 24029–24034 (2015). [CrossRef] [PubMed]
6. D. Qian, T. T. O. Kwok, N. Cvijetic, J. Hu, and T. Wang, “41.25 Gb/s real-time OFDM receiver for variable rate WDM-OFDMA-PON transmission,” in OFC (OSA, 2010), paper PDPD9.
7. X. Xiao, F. Li, J. Yu, X. Li, Y. Xia, and F. Chen, “100-Gb/s single-band real-time coherent optical DP-16QAM-OFDM transmission and reception,” in OFC (OSA, 2014), paper Th5C.6.
8. Q. W. Zhang, E. Hugues-Salas, Y. Ling, H. B. Zhang, R. P. Giddings, J. J. Zhang, M. Wang, and J. M. Tang, “Record-high and robust 17.125 Gb/s gross-rate over 25 km SSMF transmissions of real-time dual-band optical OFDM signals directly modulated by 1 GHz RSOAs,” Opt. Express 22(6), 6339–6348 (2014). [CrossRef] [PubMed]
9. M. Chen, J. He, Q. Fan, Z. Dong, and L. Chen, “Experimental demonstration of real-time high-level QAM-encoded direct-detection optical OFDM systems,” J. Lightwave Technol. 33(22), 4632–4639 (2015). [CrossRef]
10. X. Yi and K. Qiu, “Estimation and compensation of sample frequency offset in coherent optical OFDM systems,” Opt. Express 19(14), 13503–13508 (2011). [CrossRef] [PubMed]
11. H. Shafiee, B. Nourani, and M. Khoshgard, “Estimation and compensation of frequency offset in DAC/ADC clocks in OFDM systems,” in Proc. ICC (IEEE, 2004), pp. 2397–2401. [CrossRef]
12. W. Wang, Q. Zhuge, M. Morsy-Osman, Y. Gao, X. Xu, M. Chagnon, M. Qiu, M. T. Hoang, F. Zhang, R. Li, and D. V. Plant, “Decision-aided sampling frequency offset compensation for reduced-guard-interval coherent optical OFDM systems,” Opt. Express 22(22), 27553–27564 (2014). [CrossRef] [PubMed]
13. X. Q. Jin and J. M. Tang, “Optical OFDM synchronization with symbol timing offset and sampling clock offset compensation in real-time IMDD systems,” IEEE Photonics J. 3(2), 187–196 (2011). [CrossRef]
14. R. P. Giddings and J. M. Tang, “Experimental demonstration and optimisation of a synchronous clock recovery technique for real-time end-to-end optical OFDM transmission at 11.25Gb/s over 25km SSMF,” Opt. Express 19(3), 2831–2845 (2011). [CrossRef] [PubMed]
15. M. Chen, J. He, J. Tang, and L. Chen, “Pilot-aided sampling frequency offset estimation and compensation using DSP technique in DD-OOFDM systems,” Opt. Fiber Technol. 20(3), 268–273 (2014). [CrossRef]
16. R. Deng, J. He, M. Chen, and L. Chen, “SFO compensation by pilot-aided channel estimation for real-time DDO-OFDM system,” Opt. Commun. 355, 172–176 (2015). [CrossRef]
17. Z. Zhang, Q. Zhang, Y. Li, Y. Song, J. Zhang, and J. Chen, “A single pilot subcarrier-based sampling frequency offset estimation and compensation algorithm for optical IMDD OFDM systems,” IEEE Photonics J. 8(5), 7804209 (2016). [CrossRef]
18. Z. Zhang, Q. Zhang, J. Chen, Y. Li, and Y. Song, “Low-complexity joint symbol synchronization and sampling frequency offset estimation scheme for optical IMDD OFDM systems,” Opt. Express 24(12), 12577–12587 (2016). [CrossRef] [PubMed]
19. D. Chang, T. M. N. Ngatched, O. A. Dobre, R. Venkatesan, and O. Omomukuyo, “Simple sampling clock synchronisation scheme for reduced-guard-interval coherent optical OFDM systems,” Electron. Lett. 51(24), 2026–2028 (2015). [CrossRef]
20. M. Chen, Q. Chen, R. Deng, H. Zhou, Z. Zheng, J. He, and L. Chen, “Self-cancellation of sampling frequency offset in adaptively modulated IMDD-OFDM systems,” in ACP (OSA, 2016), paper ATh2D.2.
21. X. Liu and F. Buchali, “Intra-symbol frequency-domain averaging based channel estimation for coherent optical OFDM,” Opt. Express 16(26), 21944–21957 (2008). [CrossRef] [PubMed]
22. M. Chen, J. He, J. Tang, X. Wu, and L. Chen, “Real-time 10.4-Gb/s single-band optical 256/64/16QAM receiver for OFDM-PON,” IEEE Photonics Technol. Lett. 26(20), 2012–2015 (2014). [CrossRef]
23. D. Qian, M. F. Huang, E. Ip, Y. K. Huang, Y. Shao, J. Hu, and T. Wang, “High capacity/spectral efficiency 101.7-Tb/s WDM transmission using PDM-128QAM-OFDM over 165-km SSMF within C- and L-bands,” J. Lightwave Technol. 30(10), 1540–1548 (2012). [CrossRef]
24. F. Li, X. Li, L. Chen, Y. Xia, C. Ge, and Y. Chen, “High-level QAM OFDM system using DML for low-cost short reach optical communications,” IEEE Photonics Technol. Lett. 26(9), 941–944 (2014). [CrossRef]
25. L. Peng, M. Hélard, and S. Haese, “On bit-loading for discrete multi-tone transmission over short range POF systems,” J. Lightwave Technol. 31(24), 4155–4165 (2013). [CrossRef]
26. R. A. Shafik, S. Rahman, and R. Islam, “On the extended relationships among EVM, BER and SNR as performance metrics,” in Proc. ICECE (IEEE, 2006), pp. 408–411. [CrossRef]
